You are on page 1of 9

See

discussions, stats, and author profiles for this publication at:


http://www.researchgate.net/publication/7963938

A critical process analysis of wine


production to improve cost, quality
and environmental performance

ARTICLE in WATER SCIENCE & TECHNOLOGY · FEBRUARY 2005


Impact Factor: 1.11 · Source: PubMed

CITATIONS READS

4 35

4 AUTHORS, INCLUDING:

Florian F Bauer
Stellenbosch University
70 PUBLICATIONS 2,030 CITATIONS

SEE PROFILE

Leon Lorenzen
Stellenbosch University
69 PUBLICATIONS 1,160 CITATIONS

SEE PROFILE

Available from: Leon Lorenzen


Retrieved on: 28 September 2015
A critical process analysis of wine production to improve cost,

Water Science and Technology Vol 51 No 1 pp 39–46 ª IWA Publishing 2005


quality and environmental performance
C.M. Sheridan*,***, F.F. Bauer*, S. Burton** and L. Lorenzen***
* Institute for Wine Biotechnology, University of Stellenbosch, Private Bag X1, Matieland, 7602, South
Africa
** Department of Chemical Engineering, University of Cape Town, Private Bag Rondebosch, Cape Town,
7701, South Africa
*** Department of Process Engineering, University of Stellenbosch, Private Bag X1, Matieland, 7602,
South Africa

Abstract Wine production in South Africa is delocalised, with numerous small-to-medium sized
producers within several regions within the Western Cape. Whilst adapting to new technological
changes, producers have to respond to pressure from consumers and governments regarding the
environmental consequences of winemaking, especially water usage and pollution. To date, no
systematic analysis integrating the various aspects of winemaking in South Africa has been done. This
study assessed both physical inputs and outputs. A detailed questionnaire was developed to broadly
assess these parameters and was submitted to all cellars in South Africa. Case studies were performed at
three cellars during the 2002 harvest season to validate the questionnaires and collect missing
information. Based on this, and a cocurrent project, the following parameters were correlated to the tons
of grapes presses per annum: effluent parameters which include chemical oxygen demand, suspended
solids, total dissolved solids, sodium adsorption ratio, quantity of effluent; wine produced, water
consumed, and electricity consumed. These parameters were used to develop an input/output model.
This model may be used by wineries to predict their water and electrical consumption, wine produced and
effluent characteristics provided they know the tonnage of grapes pressed per year.
Keywords Input/output correlations; process analysis; wine

Introduction
Wine production generally follows traditional methodologies, however new technologies
have resulted in important changes in winemaking over the last few decades. Whilst adapting
to these technological changes, producers also have to respond to increased pressure from
consumers regarding the quality of the product and the environmental consequences of
winemaking, particularly with regard to water usage and chemical pollution. No systematic
analysis integrating all the different aspects of the winemaking process in the South African
wine industry has thus far been undertaken. This project therefore systematically analyses
the process of wine production during cellar operation, from the reception of grapes to the
final product ready for bottling. This analysis aimed to assess all of the physical inputs
(e.g. cellar infrastructure, energy, chemical and water consumption); problems associated
with processing (occurrence of microbial contamination, problems during clarification,
stuck and/or sluggish fermentation) and the output in terms of income, product quality and
the amount of effluent/waste generated.
Due to a lack of information regarding some aspects of winemaking in South Africa, this
proved difficult. A significant amount of information is available in terms of cellar
throughput, and hectares under vines, but there is little data on consumption levels of water or
electricity, what equipment is used in cellars, the frequency of microbial contamination of
the wines or stuck fermentations. 39
Duarte et al. (1998) analysed the process of wine production in Portugal. Their study
analysed parameters such as COD, BOD, TDS etc. and performed a mass balance over a
winery. However, their data was specific for one winery only. This study found that that the
daily water flow during the harvest was twice that of the first racking period, which is useful
to size a wastewater treatment plant.
Rozzi et al. (1998) tried to estimate specific polluting loads for 17 different European
wineries. The specific load (gCOD/100 kg grapes.day) ranged from 4.3 to 12. This study
C.M. Sheridan et al.

concluded that higher polluting loads occurred during the harvest period, and that the
polluting load could be correlated to the quantity of grapes harvested. The data showed,
however, that there was no clear correlation between the quantity of grapes harvested and the
polluting load during the racking season.
Balsari and Airoldi (1998) developed a software package for winery waste management.
In order for such a system to work correlations had to be developed. The data presented in
this paper showed that both the washing water and the COD of the washing water could be
correlated to the quantity of grapes pressed. Their correlations were developed from
experimental data collected at various cellars in the Cuneo area in 1995 and 1996.
As such, the objectives of this study were to:
 Develop a questionnaire for submission to wineries that could be used as a basis for
further development and data collection.
 Obtain a set of data with input from as many wineries as possible.
 Correlate various parameters with a specific winery input (grapes pressed).
 Develop a mathematical model of a South African winery that could be used as a rough
guide to predict various parameters based solely on the tons of grapes pressed per annum.

Methods
Data was obtained by the development of a questionnaire that assessed a number of broad
parameters. This questionnaire was submitted to 390 cellars in South Africa (mid 2001).
Sixty replies were finally returned and these form the basis of much of the data presented in
this thesis. Additional data was obtained from the Winetech effluent sampling programs due
to both sets of data being incomplete in certain instances. The merging of the data sets
allowed for a more meaningful analysis of the information of both databases. The data was
analysed and parameters were correlated to the tons of grapes pressed by each cellar.
Microsoft Excel 2000 was used for the analysis of the data, and the built-in line fitting
functions were used to generate equations as well as regression coefficients.

Results and discussion


Physical inputs
Physical inputs can be described as the parameters whereby materials/energy enter the
process of winemaking. Typical inputs can be defined as grapes, electricity consumption,
chemicals and water. All variables have been correlated to tons of grapes pressed per year
because this is the main material input.

Electricity
Figure 1 shows how many electrical units (kWhr) were consumed per month as a function of
the tons of grapes pressed. This data is taken from questionnaire respondents who measure
cellar electricity consumption. There are two correlations shown – during harvest and out of
harvest. During the harvest, Eq. (1) may be used, where EH is electrical units per month, and
T is the tons pressed per season. This correlation has a co-efficient of regression value of
40 0.7689, which indicates low level of data scatter.
900000

800000
Electricity Used During Harvest Season
700000
Electrical Units/month

Electricity Used Out of Harvest Season


600000

500000
0.8587
y = 88.634x
400000 2
R = 0.7689

C.M. Sheridan et al.


300000 0.811
y = 42.817x
2
200000 R = 0.5137

100000

0
0 5000 10000 15000 20000 25000
Tons of Grapes Pressed

Figure 1 Electricity consumed

Equation (2) shows the electricity consumption of the cellars outside of the harvest
season. As is expected, the consumption of electricity is lower. Harvest consumption is
approximately 2.5 times that of the rest of the year. However, the scatter of the data is greater,
so this correlation is not as accurate as the harvest data. This is possibly due to different types
of lights and different heating regimes used by cellars in winter. Most cellars use refrig-
eration in summer, which makes up a large percentage of the energy costs.
EH =88:63  T 0:8587 (1)
EO =44:82  T 0:811 (2)

Water
In Figure 2, the consumption of water per year was plotted as a function of the tons pressed
per year. The data was split into wineries with water measurement and those without.
W=4037:5  T 0:9243 (3)
Equation (3) has relatively low scatter, and can thus be considered accurate. It is inter-
esting to note from this graph that there is difference between real water measurement and
guesswork. Estimating seems to undervalue the consumption of water by approximately

50000000
45000000 Water Usage
40000000
Water Consumed/L/year

People who think they know how much water they use
35000000
30000000
y = 4037.5x0.9243
25000000
R2 = 0.8514
20000000
15000000
10000000 y = 3984.3x0.8266
5000000 R2 = 0.579
0
100 5100 10100 15100 20100 25100
Tonnage Pressed

Figure 2 Water consumption 41


60%. Given that only 20% (Sheridan, 2003) of cellars measure their water, this implies that
80% would underreport their water consumption, or would underestimate it if required to
report a value.

Physical outputs
Physical outputs may be defined as all of the products leaving the cellar. These include wine,
C.M. Sheridan et al.

solid waste, lees and wastewater. These variables were correlated to the tonnage pressed, and
the following set of graphs was developed.

Wine
In Figure 3, the quantity of wine produced was plotted as a function of tons of grapes pressed.
There is a strong correlation (Eq. (3)) between tons pressed and wine produced, as is to be
expected. Figure 3 may seem trivial, but it allows the correlation of all variables to wine
produced, instead of tons pressed. This is useful because wine is ultimately the product that is
sold. It can be noticed that there are significant variations between producers. These
variations do have a significant economic effect. It is normally expected that those wineries
that produce less wine per ton of grapes pressed (those points below the correlation line)
make better quality wines. This is not the case in this data set. All three wineries that have
data points far below the correlation line tend to make “lower quality” wine. A possible
explanation for this is that the pressing equipment is old and not as efficient as the more
modern presses.
Wine=626:24  T (4)

Effluent
Effluent is defined as the wastewater that is discharged from a cellar during normal and
abnormal operation. Abnormal operation could be defined as any problem that occurs during
processing. These could include extra busy harvest periods, the accidental release of wine
or juice etc. into the drains. For the purposes of this study, the effluent has been
quantified according to the following classifications: quantity disposed, effluent quality,
chemical oxygen demand (COD), total dissolved solids (TDS) and sodium adsorption
ratio (SAR).

10000000
y = 626.24x
9000000
R2 = 0.9437
8000000
Litres of Wine Produced

7000000

6000000

5000000

4000000

3000000

2000000

1000000

0
0 2000 4000 6000 8000 10000 12000 14000 16000
Tons Pressed

42 Figure 3 Wine produced as a function of tons of grapes pressed


Quantity of effluent
From the data gathered, only 5% of respondents measure the quantity of effluent disposed,
therefore no meaningful data on the quantity of effluent disposed could be obtained. One
cellar measured effluent disposal to be 10% more than their incoming water flow. Since this
value would oversize any treatment plant, it was used for the purposes of this study.
Effluent=1:1  Water (5)

C.M. Sheridan et al.


Effluent characterisation
In Figure 4, a plot was made to correlate various characteristics of the effluent with the tons
pressed by the cellar. It is evident from the graph that as the size of the cellar increases, the
quality the effluent decreases (the concentration of COD and TDS rises).
The COD can be correlated to the quantity of tons pressed with Eq. (6). However, this
equation has a very low co-efficient of regression which implies a high level of data scatter
and consequent poor predictability.
COD=772:2  T 0:2753 (6)
On the same figure, the TDS has also been plotted as a function of tons pressed. This
correlation has an even lower regression co-efficient. This shows that these correlations are
not particularly accurate, but they do provide industrial averages. The TDS in solution can be
given as a function of the tons pressed, shown in Eq. (7).
TDS=380:0  T 0:2081 (7)
In Figure 5, the SAR (sodium adsorption ratio: defined as
[Na]
SAR=
([Ca]+[Mg])1=2
where concentrations are given in mmol/L) of the effluent streams was plotted as a function
of the tons of grapes pressed and correlated. It must be noted that from all the data, only one
cellar exceeded the permissible limit for the SAR for irrigation of effluent (The legal limit for
irrigation of effluent is 3 in South Africa). It does show that the SAR of the effluent increases
linearly with respect to tons of grapes pressed per annum. This is given by Eq. (8).
SAR=6:10x5  T+1:0414 (8)

100000

y = 772.2x 0.2753
COD TDS R2 = 0.3231

10000
Value mg/L

1000 y = 380.03x 0.2081


R2 = 0.2574

100
100 1000 10000 100000
Tonnage Pressed

Figure 4 Effluent characterisation – COD and TDS as a function of tons of grapes pressed 43
3,5

3
y = 6E-05x + 1.0414
2,5 R2 = 0.4355

2
SAR
1,5
C.M. Sheridan et al.

0,5

0
0 5000 10000 15000 20000 25000
Tons

Figure 5 Effluent characterisation – SAR as a function of tons of grapes pressed

Development of an input/output model


Figure 6 is a pictorial representation of a preliminary black box model developed to perform
mass balance and financial balance calculations for South African wineries based solely on
the tons of grapes pressed.
An input is defined as something that is added to the process. In this case, grapes, water,
chemicals and energy are defined as the inputs. Typically, inputs have an associated cost.
Outputs are materials/energy that are removed from a process. In this case, the outputs are
designed as wine, wastewater, solid waste, lees, energy losses and evaporation losses.
Outputs may have associated costs, but may also be value added products. These however,
are not exclusive definitions. There are many more inputs and outputs, but for the purposes of
this study, the above are the most important.

Wine = Income
F8
Grapes (T)

F4

F1
F5
Waste Water
F7
Water COD

F6 TDS
F2
Black Box Winery SAR

Chemicals

Process
F3 Param ing
eters
Solid Waste
Energy Design
/Equi
Param pment
eters

Lees
Evaporation Losses

Energy Losses

44 Figure 6 Mathematical model of a winery


Processing parameters are defined as those parameters that can be changed by adjusting or
modifying the way wine is made. One example of a processing parameter would be applying
pinch technology heat exchange (Ficarella and Laforgia, 1999). Another example would be
adjusting the residence times of the drainage sumps to lower the ‘leaching’ of COD from pips
and skins into the juice.
Design or equipment parameters are defined as those parameters that are changed by
adjusting the equipment in the process. Equipment parameters may be very costly to change

C.M. Sheridan et al.


but this is not necessarily the rule. The parameters range from fixing the cracks and chips in
open fermenters, to replacing mild steel tanks with stainless steel ones.
However, there are parameters that are missing from this model that cannot be so easily
quantified, an example of this being the solid waste. No information was obtained so there is
no cost that can be assigned to the quantity of solid waste. Similarly, the lees can be sold for
tartrate recovery or dumped with the solid waste; this then alters the economics of the model.
It was unfortunately not possible to include chemical consumption in this model. Many
cellars do not measure chemical usage and there are many different types of chemicals that
may be used for the same purpose. The questionnaire sought to obtain this information, but
the replies were in many cases were either too vague or omitted. This unfortunately means
that there is a missing parameter, which is of great importance.
The missing parameters do not make the model less useful though. Any information that
is missing can be added at a later stage and the model can then be improved on. Such is the
purpose of this model; it is the first of its type and can be used by later studies or by industry.
It can also be used by cellars and fine-tuned for their purposes, according to their own unique
set of operating conditions, e.g. for red and white grapes.
The following functions have been defined in the numbered equations during this study:
F1: Water=4037.5 . T 0.9243 (3)
F2: EH=88.63 . T 0.8587 (1)
F3: EO=44.82 . T 0.811 (2)
F4: COD=772.2 . T 0.2753 (6)
F5: TDS=380.0 . T 0.2081 (7)
F6: SAR=6 . 10x5 . T+1.0414 (8)
F7: Effluent=1.1 . Water (5)
F8: Wine=626.24 . T (4)

Conclusions
It was found that although questionnaires do provide valuable information, they are quite
restrictive when seeking detailed and quantifiable data. It meant therefore, that the data
obtained from this study had to be complemented by data from other sources, in this case
from the Winetech program for ongoing effluent analysis. However, the combined data sets
provided a detailed database which can be utilised for further research by both the wine
industry and academia.
Various parameters with varying degrees of accuracy were correlated to the tons of grapes
pressed per annum and this allowed a preliminary input/output model to be developed. It is
possible to use this model to roughly predict certain cellar parameters. Two instances of these
parameters include effluent COD and TDS. Considering the lack of measurement of these
parameters in the industry, this model proves to be a useful tool when dealing with cellars
that require information on their processes but have made no measurements in the past. It
enables them to anticipate what the parameters are likely to be, and they can use these
parameters as a benchmark when comparing themselves to other cellars. This model 45
provides an industrial “average” for the abovementioned variables and it is always gratifying
to a cellar if they can establish that they use less water than the average cellar of that size.
The model needs refinement though, in terms of quantifying inputs like chemical
consumption, but also due to it being a dynamic model. As cellars change their processes to
become more environmentally friendly and efficient, the correlations will change and this
needs to be reflected in the model. It is also possible that this model bears little resemblance
to wineries in other parts of the world where winemaking philosophies and priorities differ.
C.M. Sheridan et al.

Based on these results, further research is being performed on the use of constructed
wetlands for winery effluent treatment, primarily by identifying those wetland botanical and
microbial communities best suited for winery effluent. Using the abovementioned model, the
cellars that fall within an (unspecified as yet) effluent strength range will be identified as
suitable candidates for the use of this system for treatment of their effluent.

Acknowledgements
The authors wish to thank: NRF and Winetech for sponsoring this study; Winetech for
supplying effluent sampling data; the Department of Water Affairs and Forstry for supplying
data or the “Effluent and waste disposal study in the western Cape Town of Robertson’; those
cellars that completed the questionnaire; and the three cellars that allowed the first author
onto their premises over the 2002 harvest.
The first author (C.M.S) also wishes to thank Dr. F. Baner and Professor L. Lorenzen
for their supervision of this study.

References
Balsari, P. and Airoldi, G. (1998). WIWa: a software for winery waste management. Proceedings from the 2nd
International Specialised Conference on Winery Wastewaters, Bordeaux, France. May 5–7, 1998.
Cemagref-DICOVA.
Duarte, E.A., Martins, M.B., Ghira, J.P., Carvalho, E.C., Spranger, I., Cost, S., Leandro, M.C. and Duarte, J.M.
(1998). An integrated approach for assessing the environmental impacts of wineries in Portugal.
Proceedings from the 2nd International Specialised Conference on Winery Wastewaters, Bordeaux,
France. May 5–7, 1998. Cemagref-DICOVA.
Ficarella, A. and Laforgia, D. (1999). Energy conservation in alcohol distillery with the application of pinch
technology. Energy Conservation and Management 40, 1495–1514.
Rozzi, A., Malpei, F. and Padoani, L. (1998). Estimate of polluting loads in effluents of Italian north east
wineries. Proceedings from the 2nd International Specialised Conference on Winery Wastewaters,
Bordeaux, France. May 5–7, 1998. Cemagref-DICOVA.
Sheridan, C.M. (2003). A critical process analysis of wine production to improve cost, quality and
environmental performance. MSc thesis, Institute for Wine Biotechnology, University of Stellenbosch.

46

You might also like